|Publication number||US8136025 B1|
|Application number||US 10/613,637|
|Publication date||Mar 13, 2012|
|Filing date||Jul 3, 2003|
|Priority date||Jul 3, 2003|
|Also published as||US9411889, US20120173552|
|Publication number||10613637, 613637, US 8136025 B1, US 8136025B1, US-B1-8136025, US8136025 B1, US8136025B1|
|Inventors||Huican Zhu, Anurag Acharya|
|Original Assignee||Google Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (116), Non-Patent Citations (46), Referenced by (31), Classifications (22), Legal Events (2)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present invention relates to the field of search engines for locating documents in a collection of documents, such as a collection of documents stored on servers coupled to the Internet or in an intranet, and in particular the present invention relates to a method and apparatus for assigning document identification tags to documents to be inserted into an index of the collection of documents, the index arranged and document identification tags assigned so that documents with high relevance appear earlier in a list of documents returned by the index in response to a query, even when not all the documents to be indexed are available for sorting as document identification tags are assigned and documents inserted into the index.
Development of a search engine that can index a large and diverse collection of documents, yet return to a user a short, relevant list of result documents in response to a query has long been recognized to be a difficult problem. Various metrics of relevance of a document have been developed in an attempt to solve this problem. One class of such metrics is the query-independent metrics. The metrics represent the relative importance or relevance of a document to a user independent of any query submitted. Examples of query-independent metrics include, but are not limited to, simple criteria based on intrinsic properties of the document itself (i.e., the length of the document), ad-hoc rules for assigning relevance based on preassigned authority of a hosting site, and automatic determinations of relevance based on extrinsic information about the document. An example of an automatic relevance criterion based on extrinsic information is PageRank, described in detail in U.S. Pat. No. 6,285,999, hereby incorporated by reference in its entirety.
One goal of search engine design is to index documents in such a way that a list of documents returned in response to a query is approximately ordered in decreasing relevance. This task is made easier if the list of documents is ordered in terms of decreasing query-independent relevance. For computational efficiency, it is desirable that the internal representation of documents in the index reflect such an ordering. In this way, the list of documents returned to the user will contain the most highly relevant documents (as measured by a query-independent relevance metric), even when only the first few documents in the index are returned. Extracting only the first few documents from the index has advantages in computational efficiency, a critical factor when hundreds of millions of queries are served per day.
In search engine systems that retrieve (“crawl”) and evaluate the entire contents of a collection of documents before building an index, the index is readily assembled to return documents in order of decreasing query-independent relevance. Some indexes employ an internal representation of a particular document, referred to as a document identification tag. In some systems, the document identification tags are integers. By examining the query-independent relevance of a document relative to the collection of documents prior to the assignment of a document identification tag to the document, it is possible to assign a document identification tag that encodes this information. For example, assuming sufficient computational resources, the entire collection of documents could be sorted in order of decreasing query-independent relevance and document identification tags assigned in sequential order to documents in the sorted list.
However, as the number of documents in the Internet grows, it takes ever longer time periods between the time when a page is crawled by a robot and the time that it can be indexed and made available to a search engine. Furthermore, it takes ever longer time periods to replace or update a page once it has been indexed. Therefore, what is needed in the art are systems and methods for crawling and indexing web pages to reduce the latency between the time when a web page is either posted or updated on the Internet and the time when a representation of the new or updated web page is indexed and made available to a search engine.
Given the above background, it is desirable to devise a system and method for assigning document identification tags to documents to be indexed before retrieval of the entire contents of a collection of documents is complete. Furthermore, it is desirable to devise systems and methods for assigning document identification tags before a crawl is complete in such a way that the document identification tags encode information about the query-independent relevance of the document relative to the collection of documents.
Embodiments provide a method of assigning a document identification tag to a new document. The new document is to be added to a collection of documents. First, the method subdivides a set of document identification tags into a plurality of tiers. Each tier is associated with a respective subset of the set of document identification tags. Next, query-independent information about the new document, including a query-independent metric, is received. Next, based at least on the query-independent information, one of the tiers is selected. Finally, the new document is assigned a document identification tag from the respective subset of document identification tags associated with the selected tier, the assigned document identification tag not previously assigned to any of the documents in the collection of documents.
In some embodiments, each tier in the plurality of tiers is associated with a respective predetermined range of metric values. In these embodiments, selecting a tier includes selecting the tier for which the query-independent metric of the new document falls within the respective predetermined range of metric values associated with the selected tier. In some of these embodiments, the plurality of tiers are ordered so that the respective subsets of the set of document identification tags associated with each tier monotonically increases with the position of the tier in the ordering and the respective predetermined range of metric values associated with each tier monotonically increases with the position of the tier in the ordering. In some embodiments, assigning a document identification tag to the new document includes assigning to the new document a minimum available document identification tag from the respective subset of document identification tags associated with the selected tier. In other embodiments, assigning a document identification tag to the new document includes assigning to the new document a maximum available document identification tag from the respective subset document identification tags associated with the selected tier.
In some embodiments, when a flush condition is satisfied, a flush operation is performed. The flush operation includes building a sorted map, the sorted map relating globally unique identifiers to document identification tags assigned to documents since a prior flush operation. In some embodiments, when a merge condition is satisfied, a merge operation is performed. The merge operation includes merging a plurality of sorted maps produced by previous flushing operations, producing a merged map relating globally unique identifiers to document identification tags assigned to documents.
In some embodiments, the flush operation includes building a first sorted map and a second sorted map. The first sorted map is keyed and sorted by globally unique identifiers, and includes for each globally unique identifier a corresponding document identification tag. The second sorted map is keyed and sorted by document identification tags assigned to documents since a prior flush operation, and includes for each such document identification tag a corresponding globally unique identifier.
Another aspect of the invention partitions a set of valid globally unique document identifiers into a plurality of segments, each segment associated with a respective subset of the set of valid globally unique document identifiers. Each of the segments is subdivided into a plurality of tiers, each segment having an associated set of document identification tags, and each tier of a respective segment having an associated subset of the set of document identification tags for the respective segment. Embodiments in accordance with this aspect receive query-independent information about a new document, including a query-independent metric and a globally unique document identifier. Then, based at least in part on the globally unique document identifier, one of the segments is selected. Next, based at least on the query-independent information, one of the tiers associated with the selected segment is selected. Finally, the new document is assigned a document identification tag from the subset of document identification tags associated with the selected tier. The receiving, selecting a segment, selecting a tier, and assigning is repeated with respect to one or more additional new documents. The set of document identification tags for each respective segment have the property that none of the tags therein are assigned to documents that have globally unique document identifiers associated with the respective segment.
Another aspect partitions a set of valid globally unique document identifiers into a plurality of segments, each segment associated with a respective subset of the set of valid globally unique document identifiers. Each of the plurality segments is further partitioned into a respective set of partitions, each partition associated with a respective range of the subset of globally unique document identifiers associated with the segment. Embodiments in accordance with this aspect subdivide each of the partitions into a plurality of tiers, each partition having an associated set of document identification tags, and each tier of a respective partition having an associated subset of the set of document identification tags for the respective partition. A server responsible for assigning document identification tabs receives query-independent information about a new document, including a query-independent metric and a globally unique document identifier. In one embodiment, the globally unique document identifier is a fixed length fingerprint of the documents' address or URL. Based at least in part on the globally unique document identifier, one of the segments and one of the partitions is selected. In addition, based at least on the query-independent information, one of the tiers associated with the selected partition is selected. The new document is assigned a document identification tag from a subset of document identification tags associated with the selected tier. The receiving, selecting a segment and a partition, selecting a tier, and assigning steps or operations are repeated with respect to one or more additional new documents.
Another aspect provides a system for assigning a document identification tag to a new document. The system includes a data structure representing a subdivision of a set of document identification tags into a plurality of tiers, wherein each tier is associated with a respective subset of the set of document identification tags. An interface is configured to receive query-independent information about the new document, the information including a query-independent metric. The system further includes a tag assignment module configured to (1) select, based at least on the query-independent information, one of the tiers and (2) assign to the new document a document identification tag from the respective subset of document identification tags associated with the selected tier, the assigned document identification tag not previously assigned to any of the documents in the collection of documents.
Another aspect provides a computer program product for use in conjunction with a computer system, the computer program product comprising a computer readable storage medium and a computer program mechanism therein. The computer program mechanism comprises:
a data structure representing a subdivision of a set of document identification tags into a plurality of tiers, wherein each tier is associated with a respective subset of the set of document identification tags;
an interface configured to receive query-independent information about a new document, the information including a query-independent metric; and
a tag assignment module including instructions for selecting, based at least on the query-independent information, one of the tiers, and for assigning to the new document a document identification tag from the respective subset of document identification tags associated with the selected tier, the assigned document identification tag not previously assigned to any of the documents in a collection of documents.
The aforementioned features and advantages of the invention as well as additional features and advantages thereof will be more clearly understood hereinafter as a result of a detailed description of preferred embodiments of the invention when taken in conjunction with the drawings.
Like reference numerals refer to corresponding parts throughout the several views of the drawings.
The present invention provides systems and methods for crawling and indexing web pages. Advantageously, these systems and methods reduce the latency between the time when a web page is posted or updated on the Internet and the time when a representation of the new or updated web page is indexed and made available to a search engine.
In particular, the present invention provides systems and methods for assigning document identification tags to documents to be added to an index of web pages. Advantageously, these systems and methods are capable of assigning document identification tags before all of the documents to be added to the index have been crawled, while still assigning document identification tags so that lists of documents emerge from the index approximately ordered by decreasing query-independent relevance.
Data structure for storing URLs. Referring to
Periodically (e.g., daily) one of the segments 112 is deployed for crawling purposes, as described in more detail below. In addition to segments 112, there exists a daily crawl layer 104. In one embodiment, daily crawl layer 104 comprises more than fifty million URLs. Daily crawl layer 104 comprises the URLs that are to be crawled more frequently than the URLs in segments 112. In addition, daily crawl layer 104 comprises high priority URLs that are discovered by system 200 during a current epoch.
In some embodiments, data structure 100 further comprises an optional real-time layer 106. In some embodiments, optional real-time layer 106 comprises more than five million URLs. The URLs in real-time layer 106 are those URLs that are to be crawled multiple times during a given epoch (e.g., multiple times per day). For example, in some embodiments, the URLs in optional real-time layer 106 are crawled every few minutes. Real-time layer 106 also comprises newly discovered URLs that have not been crawled but should be crawled as soon as possible.
The URLs in layers 102, 104, and 106 are all crawled by the same robots 208 (
URL Discovery. There are a number of different sources for the URLs used to populate data structure 100. One source of URLs is the direct submission of URLs by users to the search engine system. Another source of URLs is through discovery of outgoing links on crawled pages. A third source of URLs is through submissions (e.g., time-based submissions) from third parties who have agreed to provide content. For example, such third parties can give links as they are published, updated, or changed.
Automated time-based submissions are made possible using technologies such as RDF Site Summary (RSS) and Resource Description Framework (RDF). RSS is a protocol, an application of XML, that provides an open method of syndicating and aggregating Web content. Using RSS files, a data feed can be created that supplies headlines, links, and article summaries from a Web site. RDF is a syntax for specifying metadata.
Before storage in data structure 100, a URL (and the content of the corresponding page) is processed by a series of modules that are designed to ensure content uniformity and to prevent the indexing of duplicate pages. For example, one such process is a URL rewrite module. The URL rewrite module strips or rewrites commands in URL statements that would produce undesirable output. For example, if a URL includes a statement that inhibits output of a certain column, the statement is stripped from the URL. In addition to examining the syntax of specific URLs, there is a host duplicate detection module. The host duplicate detection module attempts to determine which hosts are complete duplicates of each other by examining incoming URLs.
Exemplary methods. An exemplary system 200 and method by which the URLs in data structure 100 (
Step 302. In step 302 URL scheduler 202 determines which URLs will be crawled in each epoch, and stores that information in data structure 100. Controller 201 selects a segment 112 from base layer 102 for crawling. The selected segment 112 is referred to herein as the “active segment.” Typically, at the start of each epoch, controller 201 selects a different segment 112 from base layer 102 as the active segment so that, over the course of several epochs, all the segments 112 are selected for crawling in a round-robin style.
URL scheduler 202 revises daily layer 104 and optional real-time layer 106 by moving URLs to layers 104 and 106 from base layer 102 or vice versa. The decision as to whether to add or remove URLs from daily layer 104 and real-time layer 106 is based on information in history logs 218 that indicates how frequently the content associated with the URLs is changing as well as individual URL page ranks that are set by page rankers 222. In some embodiments, the determination as to what URLs are placed in layers 104 and 106, as opposed to layer 102, is made by computing a daily score of the form:
daily score=[page rank]2*URL change frequency
The mechanism by which URL scheduler 202 obtains URL change frequency data is best understood by reviewing
A query-independent score (also called a document score) is computed for each URL by URL page rankers 222. Page rankers 222 compute a page rank for a given URL by considering not only the number of URLs that reference a given URL but also the page rank of such referencing URLs. Page rank data can be obtained from URL managers 204. A more complete explanation of the computation of page rank is found in U.S. Pat. No. 6,285,999, which is hereby incorporated by reference as background information.
URL history log 218 can contain URLs that are not found in data structure 100. For instance, the URL history log 218 may contain log records for URL's that no longer exist. The URL history log 218 may also contain log records for URL's that exist but that which the URL scheduler 202 will no longer schedule for crawling (e.g., due to a request by the website owner that the URL not be crawled, due to objectionable content, or for any other reasons).
In cases where URL scheduler 202 determines that a URL should be placed in a segment 112 of base layer 102, an effort is made to ensure that the placement of the URL into a given segment 112 of base layer 102 is random (or pseudo-random), so that the URLs to be crawled are evenly distributed (or approximately evenly distributed) over the segments. In some embodiments, the fingerprint of the URL is used to achieve the random selection of a segment 112 to place the URL. A fingerprint is, for example, a 64-bit number (or a value of some other predetermined bit length) that is generated from the corresponding URL by first normalizing the URL text (for example, converting host names to lower case) and then passing the normalized URL through a fingerprinting function that is similar to a hash function with the exception that the fingerprint function guarantees that the fingerprints are well distributed across the entire space of possible numbers. In some embodiments, the fingerprint modulus N, where N is the number of segments 112 in base layer 102 (e.g. “fingerprint modulus 12”, in the case where there are 12 segments 112 in base layer 102) is used to select the segment 112 in which to place a given URL. In some embodiments, additional rules are used to partition URLs into a segment 112 of base layer 102, daily layer 104, and real-time layer 106.
In some embodiments, it is not possible to crawl all the URLs in an active segment 112, daily layer 104, and real-time layer 106 during a given epoch. In one embodiment, this problem is addressed using two different approaches. In the first approach, a crawl score is computed for each URL in active segment 112, daily layer 104, and real-time layer 106. Only those URLs that receive a high crawl score (e.g., above a threshold value) are passed on to the next stage (URL managers 204,
In embodiments where a crawl score is computed, URL scheduler 202 determines which URLs will be crawled on the Internet during the epoch by computing a crawl score for each URL. Those URLs that receive a high crawl score (e.g., above a predefined threshold) are passed on to the next stage (URL managers 204) whereas those URLs that receive a low crawl score (e.g., below the predefined threshold) are not passed on to the next stage during the given epoch. There are many different factors that can be used to compute a crawl score including the current location of the URL (active segment 112, daily segment 104 or real-time segment 106), URL page rank, and URL crawl history. URL crawl history is obtained from URL history logs 218. Although many possible crawl scores are possible, in one embodiment the crawl score is computed as:
crawl score=[page rank]2*(change frequency)*(time since last crawl).
Additionally, many modifications to the crawl score, including cutoffs and weights, are possible. For example, the crawl score of URLs that have not been crawled in a relatively long period of time can be upweighted so that the minimum refresh time for a URL is a predetermined period of time, such as two months.
In embodiments where crawl frequency is used, URL scheduler 202 sets and refines a URL crawl frequency for each URL in data structure 100. URL crawl frequency for a given URL represents the optimum crawl frequency (or, more generally, a selected or computed crawl frequency) for a URL. The crawl frequency for URLs in daily layer 104 and real-time layer 106 will tend to be shorter than the crawl frequency of URLs in base layer 102. Crawl frequency for any given URL can range from a minute or less to a time period that is on the order of months. In one embodiment, the optimal crawl frequency for a URL is computed based on the historical change frequency of the URL and the page rank of the URL.
In addition to other responsibilities, URL scheduler 202 determines which URLs are deleted from data structure 100 and therefore dropped from system 200. URLs are removed from data structure 100 to make room for new URLs that are to be added to data structure 100. In some embodiments, a “keep score” is computed for each URL in data structure 200. The URLs are then sorted by this “keep score” and URLs that receive a low “keep score” are eliminated as newly discovered URLs are added to data structure 100. In some embodiments, the “keep score” is the page rank of a URL that is determined by page rankers 222 (
Step 304. In step 304, URL managers 204 receive the active segment as well as layers 104 and 106 from URL scheduler 202. In typical embodiments, because of the computational demands imposed upon URL managers 204, each manager 204 is resident on its own dedicated server. Further, in some embodiments, real-time layer 106 is managed by a separate URL manager 204 that holds all or a substantial portion of layer 106 in high speed random access memory. The active segment and daily layers are partitioned into the remaining URL managers 204. Typically, this partitioning is performed using a modulo function or similar function on the fingerprint values (or a portion of a fingerprint value) derived from each URL in the active segment and daily layers so as to partition these URLs into a set of approximately equal sets (partitions). Each of these sets is assigned to a different URL manager 204 of a plurality of URL managers 204.
In some embodiments, data structure 100 is partitioned and stored in a plurality of servers. In such embodiments, this plurality of servers is represented by URL scheduler 202. URLs that are copied from the URL scheduler 202 servers are distributed to the servers hosting the URL managers 204 on a random basis. Further, the number of servers used by URL scheduler 202 and the number of servers that host URL managers 204 are constrained such that they are relatively primed. That is, there is no common divider between (i) the number of servers used by URL scheduler 202 and (ii) the number of servers hosting URL managers 204. One example of a relatively primed topology is the case in which URL scheduler 202 is partitioned across 11 servers and there are 13 servers for the URL managers 204.
When the number of servers between two stages of system 200 (
As discussed above, in some embodiments, real-time layer 106 is managed by a separate URL manager 204 that holds all or a substantial portion of the real-time layer in high speed random access memory. Thus, in some embodiments, the (i) number of servers that host URLs other than those from real-time layer 106 and (ii) number of servers that host portions of data structure 100, other than real-time layer 106, are relatively primed (e.g., 11 and 13).
The use of constrained numbers of servers is used in various stages of system 200 (
The storage of URLs in hash tables 600 on each server hosted by a URL manager 204 is advantageous because it provides a way of quickly accessing URL state information. For example, to obtain state information for a particular URL, all that is required is to look up the record having the hash value that corresponds to the hash of the URL. Such a lookup process is more efficient than searching through records of all the URLs held by all the URL managers 204 for a desired URL.
Representative URL state information stored in hash tables 600 includes the URL's fingerprint (called the URL fingerprint) 604, URL page rank 606, and the layer (102, 104, or 106) to which the URL belongs. In some embodiments, URL page rank 606 is not stored in hash table 600 but is stored in a data structure that is available to each URL manager 204. A wide range of additional URL state information can be stored in hash table, including information that is obtained from or derived from status logs, history logs, and page rankers. Representative state information that can be stored in hash tables 600 is described below.
Each of the URL managers 204, other than the URL manager 204 that hosts real-time layer 106, perform a variety of functions. For instance, they scan link logs 214 to discover new URLs. Link logs 214 comprise all the links that are found on scanned web pages during the current epoch. URLs that have been seen by system 200 before but have not been scheduled for crawl during the current epoch are ignored. Newly discovered URLs are added to a hash table 600 of a URL manager 204. The hash table data structure provides an advantageous mechanism for quickly determining whether a table 600 contains a URL discovered in a link log. Rather than scanning large lists of URLs to determine whether a URL in a link log is new, the URL from the link log is simply hashed and a search is made for the resultant hash value in each hash table 600. If a URL is found in a link log 214 and is not in any hash table 600, it is added to the hash table 600 of one of the URL managers 204.
In some embodiments, the number of URLs hosted by URL managers 204 exceeds the number of URLs than can be crawled during a given epoch and/or there is a risk that URLs hosted by URL managers 204 will be crawled on an infrequent basis during a given epoch. In such embodiments, the URL status information maintained for URLs by URL managers 204 can be used to ensure that, to the extent possible, the URLs matching select criteria are given high priority for crawling. In other words, URL state information can be used to prioritize which URLs will be sent to URL server 206. Several different factors can be used to accomplish this prioritization, such as URL crawl interval 608 and URL page rank 606, to name a few. URL managers 204 obtain the page rank of URLs from page rankers 222. Page rankers 222 maintain a persistent record of the page rank of URLs and actively update the page rank of URLs using link maps 220 as described in more detail below. Crawl interval 608 represents a target frequency that a URL should be crawled. For example, if a URL has a crawl interval 608 of two hours, the URL manager will attempt to crawl the URL every two hours. Any number of criteria to be used to prioritize which URLs will be delivered upon request to URL server 206, including “URL characteristics” 612 such as the category of the URL. Representative URL categories include, but are not limited to news URLs, international URLs, language categories (e.g., French, German, Japanese, etc.), and file type categories (e.g., postscript, powerpoint, pdf, html). The URL characteristics 612 for a URL may identify a plurality of URL categories to which the URL belongs.
Step 306. Periodically, URL server 206 makes requests from URL managers 204 for URLs. In response, URL managers 204 provide URL server 206 with URLs. In some embodiments, URL server 206 requests specific types of URLs from URL managers 204 based on a policy (e.g., eighty percent foreign URLs/twenty percent news URLs) that URL server 206 is enforcing. URL managers 204 are able to service such requests because of the URL state information they store for each URL in hash tables 600. Additionally, URL server 206 attempts to ensure that each URL manager 204 contributes URL requests.
URL server 206 distributes URLs from URL managers 204 to robots 208 to be crawled. Conceptually, a robot 208 is a program that automatically traverses the Web's hypertext structure by retrieving a document at a URL, and recursively retrieving all documents that are referenced by the retrieved document. The term “recursive” as used here is not limited to any specific traversal algorithm. However, in a system that retrieves and indexes billions of documents, this simple recursive methodology is not workable. Instead, each robot 208 crawls the documents assigned to it by the URL server 206. The robot passes retrieved documents to the content filters 210, which process the links in the downloaded pages, from which the URL scheduler 202 determines which pages are to be crawled. Robots 208 are unlike normal web browsers, such as Internet Explorer (Microsoft, Redmond Wash.). For instance, when a robot retrieves the document at a URL, it does not automatically retrieve content (e.g., images) embedded in the document through the use of object or other tags. Also, in one embodiment, the robots are configured to not follow “permanent redirects”. Thus, when a robot encounters a URL that is permanently redirected to another URL, the robot does not automatically retrieve the document at the target address of the permanent redirect.
In some instances, URL server 206 avoids overloading any particular target server (not shown) that is accessed by the robots 208. The URL server 206 determines the maximum number of URL requests to be sent to any particular host. It does this by making a procedure call to a server called the host load server (not shown). The host load server stores information for each known host server (i.e., a server storing documents known to the search engine) indicating the maximum request load to be imposed by entire search engine on the host server number, and the portion of that load which is currently in use or reserved by robots. The URL server 206 sends a load reservation request to the host load server, requesting the right to send download requests to a specified host server, and receives back an indication of the number of download requests that the URL server can allocate to a robot. In other words, the URL server will be told how many URLs the URL server can send to a robot for downloading. Then, URL server 206 parcels out the proper number of URLs to robots in accordance with the load reservation granted to the URL server 206 by the host load server. The robots 208 take these URLs and download (or at least attempts to download) the documents at those URLs. When URL server 206 runs out of URLs to process, it requests more URLs from URL managers 204. Furthermore, when a robot completes the process of downloading the set of URLs it received from the URL server, the host load reservations made by the URL server are released. Alternately, host load reservations are made for a fixed period of time, and are automatically released upon the expiration of that fixed period of time.
Step 308. In step 308, a plurality of robots 208 crawl URLs that are provided to the robots 208 by URL server 206. In some embodiments, robots 208 use a calling process that requires domain name system (DNS) resolution. DNS resolution is the process by which host names (URLs) are resolved into their Internet Protocol (IP) addresses using a database that provides a mapping between host names (URLs) and IP addresses. In some embodiments, enhancements to known DNS resolution schemes are provided in order to prevent DNS resolution from becoming a bottleneck to the web crawling process, in which hundreds of millions of URLs must be resolved in a matter of hours. One of these enhancements is the use of a dedicated local database 250 (
Robots 208 use various protocols to download pages associated with URLs (e.g., HTTP, HTTPS, gopher, File Transfer Protocol, etc.). Robots 208 do not follow permanent redirects that are found at URLs that they have been requested to crawl. Rather they send the source and target (i.e., redirect) URLs of the redirect to the content filters 210. Referring to
Step 310. Pages obtained from URLs that have been crawled by robots 208 are delivered to the content filters 210. In typical embodiments, there is more than one content filter 210 in system 200 because of the computational demands of the content filter 210. In step 310, content filter 210 sends information about each retrieved page to DupServer 224 to determine if the document is a duplicate of other pages. In one embodiment, the information sent to the DupServer 224 about each page includes the URL fingerprint of the page, the content fingerprint of the page, the page's page rank, and an indicator as to whether the page is source for a temporary or permanent redirect. When a duplicate is found, the page rankings of the duplicate pages (at other URLs) are compared and the “canonical” page for the set of duplicate pages is identified. If the page presented to the DupServer 224 is not the canonical page (of the set of duplicate pages), the content filter 210 does not forward the page (to the respective RTlog 226, 228, 230) for indexing. Rather, the content filter 210 makes an entry for the page in the history log 218, creates or updates an entry for the URL in the status log 212, and then ceases work on the page. In effect, a non-canonical page is deleted from the search engine, except for the entries in the history log and status log. In addition to identifying duplicate web pages, DupServer 224 assists in the handling of both temporary and permanent redirects encountered by the robots 208.
Examples of stages where the number of servers used to host the stage is constrained have been described. For, example, the number of servers used to host data structure 100 is constrained relative to the number of servers used to host URL managers 204 such that they are relatively primed. However, there are examples in system 200 in which the number of servers used to host a stage is not constrained such that it is relatively primed with respect to the number of servers used to host a prior or subsequent stage. The number of servers used to host content filters 210 represents one such example. In other words, the number of servers used to host content filters 210 is not constrained such that it is relatively primed with respect to the number of robots 208. In fact, in some embodiments, the number of servers used to host content filters 210 is a multiple of the number of servers used to host robots 208.
Step 312. In the embodiment illustrated in FIGS. 2 and 3A-3B, the content filters write out four types of log files, link logs 214, RTlogs (226, 228, or 230), history logs 218, and status logs 212. With the exception of those URLs that have been flagged as not being canonical pages (i.e., not suitable for indexing) by the DupServer 224, URLs that have been crawled by robots 208 are processed as described below. For those URLs that have been flagged as not suitable for indexing, content filter 210 will insert corresponding records in all RTlogs, the appropriate link log 214, and the history logs 218.
Step 314. In step 314, indexers 232, 240 and 242 obtain documents from the RTlogs on a high throughput basis, and generate indices for those documents. When the indices are provided to the servers of the front-end querying system (not shown), these documents become searchable by the users of the front-end querying system.
Step 316. In step 316, global state manager 216 reads link logs 214 and uses the information in the log files to create link maps 220 and anchor maps 238. Link maps 220 are keyed by the fingerprints of the source URLs in the link logs (i.e., the URLs that respectively correspond to each record 502). The records in link map 220 are similar to records 502 in link log 214 with the exception that text 506 is stripped and the records are keyed by the fingerprint of the normalized value of the source URL. Link maps are used by page rankers 222 to adjust the page rank of URLs within data structure 100. Such page rankings persists between epochs.
In addition to creating link maps 220, global state manager 216 creates anchor maps 238. In contrast to records in a link map 220, records in an anchor map 238 are keyed by the fingerprints of outbound URLs 504 present in link log 214 (
Segments and partitions for crawling. Returning attention to the division of URLs used to schedule and perform crawling of the World Wide Web,
Segment ID=URL fingerprint modulo Q.
Each segment 902 in layer 900 may be further subdivided into one or more partitions. For example, in
Even when layer 900 is daily crawl layer 104 or real-time layer 106 and the number of segments 902 is one (Q=1), multiple partitions 902-1 are sometimes used. Segmenting and partitioning the URLs in layer 900 guides the tasks of parallelizing and scheduling the crawling and indexing of the URLs in layer 900. Partitioning and segmenting, however, are not essential to the teachings of the present disclosure.
Indexing crawled URLs. After one or more of the URLs in layer 900 have been crawled, and any non-canonical duplicate documents have been filtered out, the downloaded documents (which are stored in a log file, called the RTLog) are ready to be indexed. Downloaded documents are indexed shortly after each document is downloaded. The indexing of documents proceeds while other documents are being downloaded by the robot crawlers. Furthermore, as explained in more detail below, successful crawling of a URL (actual retrieval of the document that resides at the URL) is not necessary before indexing is performed.
Indexer 232 is, at least conceptually, part of the front-end 1050 of a web search engine as well as being part of the back-end 1000 of the search engine. However, the “front end” portion of the indexer 232 will typically be implemented on a different server than the back-end server that indexes newly crawled documents. The front end portion of the indexer utilizes the indices and other data structures (e.g., DocID to fingerprint maps) generated by the indexer 232 while processing search engine queries. Front-end 1050 receives a query 1052, generally submitted from a user of a client computer (not shown) over a network connection of some type, and returns a list 1060 of documents relevant to the query. List 1060 is preferably ordered in decreasing relevance, documents near the beginning of list 1060 being the most highly relevant to the query 1052. List 1060 optionally includes, for each document on the list, a respective URL, a respective query-dependent relevance score, a respective abstract of the contents of the document, or any combination of these or other items. To generate list 1060, front-end 1050 includes a query handler 1054 and an indexer 232. After receiving query 1052, query handler 1054 submits one or more query terms 1056 to indexer 232. In a front-end 1050 to a system that crawls more than one type of layer (for example, various combinations of a base layer, a daily layer, or a real-time layer), query terms 1056 are submitted to each respective indexer 232 for each layer, and the responses are merged. Indexer 232 returns a list 1058 of URLs that refer to documents containing one or more of the query terms 1056. In a preferred embodiment, list 1058 comprises URL fingerprints of such documents. Advantages of representing the documents in list 1058 as URL fingerprints include the storage savings afforded by use of a fingerprint, which is preferably a 64-bit integer, rather than use of the full URL (which may vary considerably in length) for a given document. Upon receiving list 1058 (or, possibly, receiving a separate list 1058 from multiple indexers 232, 240, and 242), query handler 1054 performs processing that may include merging of multiple lists 1058 and determination of a query-specific relevance score for one or more documents on lists 1058. Finally, query handler returns a list 1060, usually comprised of URLs sorted in order of decreasing query-dependent relevance, to the user. The merging and sorting of documents from lists 1058 can be accelerated if indexer 232 were to supply a list 1058 of documents sorted approximately in order of decreasing relevance. Were lists 1058 so sorted, merging multiple lists 1058 could, in some cases, consist of no more than selecting the top entry from each such list. Even if indexer 1058 were to return a list 1058 of documents sorted approximately in order of decreasing query-independent relevance, the merging and sorting performed by query handler 1054 can be accelerated. An advantage of the teachings of the present disclosure is the ability to assign document identification tags (DocIDs) to documents as they are crawled, the document identification tags encoding information about the relative query-independent relevances of the documents.
Indexer 232 is also part of back-end 1000 of a web search engine. URL server 206 submits a list of URLs to be crawled to robots 208. Robots 208, in accordance with the description of
Content filters 210 generate, based in part on information received from robots 208, one or more Rtlog files 226. In a preferred embodiment, a separate set of Rtlog files 226 is kept for each layer in the crawl: a real-time Rtlog 226, a daily Rtlog 228, and a base Rtlog 230. For simplicity, hereafter the term Rtlog 226 will be used to refer to any one of these. Not every URL passed to content filters 210 by robots 208 is necessarily included in Rtlog files 226. For example, objectionable content (pornography, slanderous sites, etc.) may be prevented from entering Rtlog files 226 by content filters 210, so that such content will not be indexed and never returned to a user in response to a query. Furthermore, there may be pages on the web which are duplicated, the same content residing at a plurality of distinct URLs. This may be the result of deliberate efforts to deceive the web search engine, in the hopes that the search engine will assign greater importance to pages so duplicated. Thus, in some embodiments, content filters 210 include or interact with a separate module for duplicate document detection and removal. Such a module is described in a related application, entitled Duplicate Document Detection in a Web Crawler System, filed on the same date as this application, hereby incorporated by reference in its entirety.
Indexer 232 examines Rtlog files 226 periodically, continuously, or in any one of a number of time fashions. Indexer 232, upon finding a URL whose content has changed since indexing was last performed, or a URL that does not appear in the index at all, retrieves information about the URL from Rtlog 226. Then, indexer 232 incorporates the most recent information about the URL in latest version of the index (not shown). In this way, recent information about the URLs that have been processed by content filters 210 is available to the front-end 1050. Thus, users submitting queries 1052 need not wait for the entire web (or even an entire segment or partition thereof) to be crawled before a document can be returned as part of the list 1060 in response to their queries. Advantages of this feature include the enhanced fidelity of the index and its responses 1058 with respect to pages whose content, location, availability, or other characteristics change rapidly (on the time-scale of days or weeks).
An Indexer for real-time indexing of documents. Examining the operation of indexer 232 in more detail,
Indexer 232 receives record 510-1 and processes it with tag assignment module 1102. Among other functions, tag assignment module 1102 determines if the document represented by record 510-1 has already been indexed and placed in index 1110. Associated with each document in the index 1110 is a document identification tag (hereafter, “DocID”) used for representing the document in the index 1110. Preferably, the DocID is a 32-bit integer. By using DocIDs to represent documents rather than, for example, the URL fingerprint of a document, the amount of storage required by index 1110 is reduced. Additionally, the URL fingerprint is determined by a hash function that does not take as part of its input any metric of relevance, query-independent or otherwise. In contrast, DocIDs are determined by module 1102 based, at least in part, on the query-independent relevance metric 514-1 of the document. Thus, a DocID can encode information about the query-independent relevance of the document to which it is assigned. In a preferred embodiment, the relative ranking of two documents in index 1110 can be determined based on the respective DocIDs of the two documents. Thus, if the DocID assigned to a first document is greater in value than the DocID assigned to a second document, it can be determined that the query-independent relevance of the second document is higher than that of the first. In other words, low DocID values correspond generally to high query-independent relevance. However, small differences in DocID values, such as between two DocID's in the same tier of DocIDs, may not indicate any difference in query-independent relevance.
Indexer 232, as a consequence of the index building process, returns inverse posting lists of documents matching a query term in order of increasing DocID. Thus, in a preferred embodiment, the inverse posting lists returned by indexer 232 are sorted in order of decreasing query-independent relevance, highly relevant documents appearing earlier in the list than less relevant documents. One goal of tag assignment module 1102 is, then, to assign DocIDs so that the DocIDs assigned approximately reflect the query-independent relevance of the associated documents.
To determine if the document represented by record 510-1 has already been placed in index 1110, tag assignment module 1102 first examines layered set 1107 of sorted fingerprint to DocID maps 1106. Layered set 1107 of sorted maps 1106 is a time-ordered sequence of sorted maps 1106. Each map (for example, 1106-1) contains one or more pairings of a URL fingerprint to a DocID. The pairings in maps 1106 are keyed and sorted by fingerprint. Each sorted map in layered set 1107 of sorted maps 1106 covers a range of logical or physical time. If the URL fingerprint 516-1 of the document represented by record 510-1 is found in any of the maps 1106 in layered set 1107, module 1102 retrieves the DocID corresponding to URL fingerprint 516-1. If, on the other hand, the URL fingerprint 516-1 of the document represented by record 510-1 is not found in any of the maps 1106 in layered set 1107, module 1102 proceeds to assign the document a new DocID.
Tag assignment module 1102 stores an internal representation of assigned DocIDs, and at various times outputs a new sorted fingerprint (FP)-to-DocID map 1106. For example, the internal representation of module 1102 may have no more room to store new DocID assignments, forcing the output of a new map 1106. In other cases, a predetermined amount of time may have lapsed since the most recent map 1106 was output, triggering the output of a new FP-to-DocID map (for example, 1106-N) so as to keep the information embodied in the layered set 1107 of maps 1106 recent. The repeated generation of maps 1106 by module 1102 results in a plurality of maps 1106 in set 1107, some of which are older than others. For example, map 1106-N may be the most recent map, containing information that supercedes, or contradicts, information in older map 1106-3. Thus, a mechanism for resolving such conflicts is needed.
A merging rule specifies how older and newer entries in the maps 1106 having the same key should be combined. In some embodiments, the merging rule dictates that newer data should completely replace older data. In other embodiments, the merging rule may be more complex. For example, a special delete entry may allow a newer entry to delete any and all older entries with the same key. The current state of all information for a particular key can be reconstructed by performing lookups in all of the sorted maps 1106 and applying the appropriate merging rule for all entries that are found corresponding to the particular key. To increase the efficiency of these lookups, subsets of the sorted maps are merged together, generating merged sorted maps. This process is described in further detail in conjunction with the discussion of
If URL fingerprint 516-1 cannot be found in the layered set 1107 of sorted fingerprint-to-DocID maps 1106, the document represented by record 510-1 is assigned a DocID before it is indexed by index builder 1108. The details of the assignment of a DocID to a new document are discussed in conjunction with
Additionally, after assigning a DocID, module 1102 provides record 510-1, along with the assigned DocID, to index builder 1108. Index builder 1108 then, when rebuilding (or adding information to) index 1110, adds the document represented by records so received to the new version of the index 1110.
The front-end aspects of indexer 232 include the ability of the index 1110 to respond to queries submitted by users or other parts of the search engine system. Index 1110 responds to query 1056 by generating a list 1111 of DocIDs. List 1111 is also referred to as an inverse posting list, as will be appreciated by one skilled in the art of database system design. For each DocID on list 1111, the respective document it represents contains at least one of the words in the query 1056. However, because the search engine system does not use DocIDs to represent documents, the DocIDs on list 1111 must be translated into URL fingerprints, or some other globally unique representation, before being output as part of response 1058 of indexer 232 to query 1058. For example, DocIDs are, in some embodiments, only unique within a particular partition or segment of a layer 900 (
Layered set 1107, only used to determine whether a particular URL FP has already been indexed, is no longer needed once indexing is complete. Set 1103, on the other hand, is a persistent data structure. Methods and data structures for compressing the size of and increasing the access speed to sorted maps 1104 in layered set 1103 are discussed in conjunction with the discussion of
Assigning a new DocID to a document.
In another embodiment, the tier records 1206 do not have a tier column and have only one of the minimum and maximum DocID columns. In this embodiment the tier number or ID is implied from the position of the record 1206 for each tier, and the other end of the DocID range is determined by inspection of the record before or after the record for the tier in question.
In yet other embodiments, the tier records 1206 include one or more additional fields to specify the range of PageRank values associated with each tier. This information is used by the Metric to Tier Lookup 1204 to select the tier for each new document, as further described elsewhere in this document.
For a URL fingerprint 516-1 that has not yet been assigned a DocID, parser 1201 first determines, by querying metric to tier lookup 1204, from which tier to assign a new DocID. Parser 1201 submits the query-independent metric 514-1 for record 510-1 and determines a selected tier. In some embodiments, metric to tier lookup 1204 is a function rather than a table. In a preferred embodiment, metric to tier lookup 1204 associates a respective range of values of a query-independent relevance metric (e.g., PageRank) with each tier. In this embodiment, parser 1202 selects a tier by selecting the tier for which the query-independent metric 514-1 of the new document falls within the respective predetermined range of metric values associated with the tier. Furthermore, in a preferred embodiment, the ranges of DocIDs associated with the tiers monotonically decrease with respect to the ranges of metric values associated with the tier by lookup 1204. That is to say that if a first tier has associated with it a first range of metric values, the first range is uniformly greater than a second range of metric values associated with a second tier, and the range of DocIDs associated with the first tier will be uniformly less than the range of DocIDs associated with the second tier.
After determining the tier from which to assign a DocID, parser 1202 examines the relevant record 1206 in data structure 1205 to determine the next available DocID for the selected tier. In
Flushing operations. Flusher 1212 monitors buffer data structure 1209 to determine if a flush condition has been met. The flush condition may depend on any number of temporal or storage considerations. For example, in one embodiment, the flush condition is satisfied if a predetermined amount of time has lapsed since the last satisfaction of the flush condition. In some embodiments, the flush condition may be satisfied if a predetermined number of mapping entries 1208 have been written to the buffer 1209. When the flush condition has been met, flusher 1212 generates two sorted maps 1104-M and 1106-N representing the mapping of DocIDs to URL fingerprints and the mapping of URL fingerprints to DocIDs, respectively. A sorted map 1104-M, keyed and sorted by DocIDs and containing one or more DocID-FP pairings, is generated. Additionally, a sorted map 1106-N, keyed and sorted by URL fingerprint, and containing one or more FP-DocID pairings, is also generated. Flusher 1212 also empties or re-initializes the buffer data structure 1209 to prepare it to receive new mapping entries.
Although depicted separately in
In other embodiments, globally unique identifiers (for example, the URLs themselves) other than URL fingerprints are stored in the slots of buffer data structure 1209.
Multi-level maps for fast resolution of URL fingerprint from DocID. As mentioned previously, the layered sorted maps 1106 that are keyed and sorted by URL fingerprint are only needed during a crawl to determine if a given URL fingerprint already has been assigned a DocID. Thus, maps 1106 may be destroyed at the end of a crawling epoch. In contrast, layered sorted maps 1104 (keyed and sorted by DocID) are needed for the front-end functionality of the web search engine. Thus, maps 1104 are persistent data structures, and it is desirable to devise methods and data structures for reducing the amount of storage and time associated with them.
To this end,
Use of multi-level map 1104-M would then consist of following the pointer in record 1402-2 to the first record 1406-2-1 of bucket 1404-2. Each record 1406-2-1, 1406-2-2, . . . , 1406-2-K in bucket 1406-2 includes two entries: a Δ DocID value, representing the difference between the DocID and the first DocID stored in the corresponding record 1402-2 in first level 1401 and the URL fingerprint associated with the DocID. By comparing the difference between the particular DocID and the first DocID 1402-2 to the Δ DocID entries in each record in bucket 1404-2, the record 1406-2-X that corresponds to the particular DocID is readily found. The URL fingerprint that the particular DocID is assigned to can then be found by reading the second entry in record 1406-2-X. In some embodiments, records 1406-2-1, 1406-2-2, . . . , 1406-2-K are sorted by Δ DocID, so that an efficient search (e.g., a bisection search) can be used to rapidly locate the appropriate record. In other embodiments, the records 1406-2-1, 1406-2-2, . . . , 1406-2-K are in no particular order. Given the relatively small size of the buckets, however, even an exhaustive search for a particular DocID is not prohibitively expensive.
Overflow of tier.
Step 1704 then checks to see if the tier is full, for instance by checking table 1205 to see if all the DocIDs allocated to the selected tier have already been assigned to other documents. If so (1704, Yes), the method selects the tier closest to the originally selected tier that is not full at step 1706. It is important to select a closest tier because the DocID assigned to each document is used by the search engine as general indication of the document's importance, and therefore the DocID assigned the document should be as close as possible to the range of DocIDs normally assigned to documents having approximately the same page rank as the new document being processed here. Once a non-full tier has been selected, the next available DocID for the selected tier is assigned to the new document (step 1708). In some embodiments, the DocID is determined by accessing data structure 1205 and looking up the “next” DocID for the selected tier, assigning that DocID to the new document, and then incrementing or decrementing the “next” DocID for the selected tier, as described above. Finally, the assigned DocID is associated with the new document by storing an URL fingerprint to DocID mapping entry in the buffer data structure at step 1710.
Merging of multiple layered sorted maps.
To mitigate this problem, tag assignment module 1102 (
To explain the merging operation in more detail, consider the set of DocID to fingerprint maps 1103. Merged map 1104-(M+1) contains all of the records from each map 1104 in the set of maps 1103. Furthermore, the records in merged map 1104-(M+1) are sorted and keyed by DocID. Thus, a rapid search algorithm (e.g., binary search) can be applied to locate the record in map 1104-(M+1) corresponding to a particular DocID. To the extent that multiple records are found in the set 1103 of maps at the time for merging, a particular merging rule that specifies how older and newer records with the same key should be combined is employed. In one embodiment, newer data completely replaces older data. In other embodiments, the merging rules are more complex. For example, a special delete entry allows a newer entry to delete any and all older entries with the same key. Notice however, that in operation of the tag assignment module 1102 in accordance with one embodiment, duplicate records such as this will not be produced in set 1103 (nor in set 1107).
Although, as depicted in
A computer system for assigning document identification tags. In a preferred embodiment, the search engine system 1800 is implemented using one or more computer systems, as schematically shown in
The computer system will typically have one or more central processing units (CPU's) 1802, one or more network or other communications interfaces 1804, primary and secondary storage 1808, and one or more communication busses 1806 for interconnecting these components. Primary and secondary storage 1808 can include high speed random access memory and can also include non-volatile memory, such as one or more magnetic disk storage devices (not shown). Primary and secondary storage 1808 can include mass storage that is remotely located from the central processing unit(s) 1802. The primary and secondary storage 1808 preferably stores:
an operating system 1810 that includes procedures for handling various basic system services and for performing hardware dependent tasks;
a network interface module 1812 that is used for connecting the system 1800 to various other computers (the content filters 210 and query handler 1054 in
a tag assignment module 1102, preferably including instructions 1814 for detection of the satisfaction of a flush condition and execution of a flush operation and instructions 1816 for detection of the satisfaction of a merge condition and execution of a merge operation;
a query-independent metric to tier lookup data structure 1204 or function for determining, based on the value of a query-independent metric for a document, from which one of the plurality of tiers a DocID will be assigned;
a data structure 1206 representing a subdivision of a set of available DocIDs into a plurality of tiers, the data structure 1206 including a respective record 1206-1, 1206-2, . . . for each tier, the record containing information about a range of DocIDs associated with the tier (embodied in the MinDocID and MacDocID columns in
a buffer data structure 1209 for temporarily storing the assignment of DocIDs to documents since a last flushing operation, the data structure including a plurality of entries 1208 storing URL fingerprint to DocID mapping information.
Tag assignment module 1102 can include executable procedures, sub-modules, tables and other data structures. Tag assignment module 1102 preferably includes instructions for receiving, via network interface module 1812 query-independent information about a new document. The query-independent information preferably includes a globally unique document identifier (such as the URL fingerprint) and a query-independent metric (such as, for example, the PageRank) rating the relevance of the document. Based on the query-independent metric, module 1102 determines, by interacting with query-independent metric to tier lookup 1204, a selected tier. Once the tag assignment module 1102 determines the selected tier, it accesses data structure 1205 to determine the next available DocID for the selected tier by accessing the appropriate record for the selected tier. Module 1102 then assigns the next available DocID as determined to the new document and, in one embodiment, increments the entry in the “Next” column of the record in data structure 1206 corresponding to the selected tier. In an alternate embodiment, module 1102 decrements this entry. The assignment of a DocID to a new document is stored in an entry 1208 in the buffer data structure 1209.
In some embodiments, tag assignment module 1102 includes instructions 1814 for detecting the satisfaction of a flushing condition and executing a flush operation. The flush condition may depend on any number of temporal or storage considerations. For example, in one embodiment, the flush condition is satisfied if a predetermined amount of time has lapsed since the last satisfaction of the flush condition. In some embodiments, the flush condition may be satisfied if a predetermined number of entries 1208 have been stored in the buffer data structure 1209.
When the flush condition has been met, tag assignment module 1102 generates two sorted maps representing the assignment of DocIDs to documents. A sorted map 1104-M (not shown) keyed and sorted by DocID is generated and distributed via network interface module 1812. Additionally, a sorted map 1106-N (not shown) keyed and sorted by URL fingerprint is generated and distributed via network interface module 1812. Additionally tag assignment module 1102 clears or re-initializes the buffer data structure 1209.
In some embodiments, tag assignment module 1102 includes instructions 1816 for detecting the satisfaction of a merge condition and for executing a merge operation. Detailed exposition of the detection of satisfaction of merge conditions and execution of merge operations can be found in the discussion of
All references cited herein are incorporated herein by reference in their entirety and for all purposes to the same extent as if each individual publication or patent or patent application was specifically and individually indicated to be incorporated by reference in its entirety for all purposes.
The foregoing description, for purposes of explanation, used specific nomenclature to provide a thorough understanding of the invention. However, it will be apparent to one skilled in the art that the specific details are not required in order to practice the invention. The embodiments were chosen and described in order to best explain the principles of the invention and its practical applications, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated. Thus, the foregoing disclosure is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings.
It is intended that the scope of the invention be defined by the following claims and their equivalents.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5634062||Oct 20, 1994||May 27, 1997||Fuji Xerox Co., Ltd.||System for managing hypertext node information and link information|
|US5721914||Sep 14, 1995||Feb 24, 1998||Mci Corporation||System and method for hierarchical data distribution|
|US5801702||Mar 9, 1995||Sep 1, 1998||Terrabyte Technology||System and method for adding network links in a displayed hierarchy|
|US5884031||Oct 1, 1996||Mar 16, 1999||Pipe Dream, Inc.||Method for connecting client systems into a broadcast network|
|US5913026||Apr 24, 1997||Jun 15, 1999||Sarnoff Corporation||Method of allocating multiple users to a multiple user data distribution system|
|US5920701||Aug 28, 1996||Jul 6, 1999||Starburst Communications Corporation||Scheduling data transmission|
|US6070191||Oct 17, 1997||May 30, 2000||Lucent Technologies Inc.||Data distribution techniques for load-balanced fault-tolerant web access|
|US6098078||Dec 23, 1996||Aug 1, 2000||Lucent Technologies Inc.||Maintaining consistency of database replicas|
|US6119124||Mar 26, 1998||Sep 12, 2000||Digital Equipment Corporation||Method for clustering closely resembling data objects|
|US6134596||Sep 18, 1997||Oct 17, 2000||Microsoft Corporation||Continuous media file server system and method for scheduling network resources to play multiple files having different data transmission rates|
|US6141762||Aug 3, 1998||Oct 31, 2000||Nicol; Christopher J.||Power reduction in a multiprocessor digital signal processor based on processor load|
|US6163778||Feb 6, 1998||Dec 19, 2000||Sun Microsystems, Inc.||Probabilistic web link viability marker and web page ratings|
|US6189019||Aug 14, 1996||Feb 13, 2001||Microsoft Corporation||Computer system and computer-implemented process for presenting document connectivity|
|US6189039||Apr 10, 1997||Feb 13, 2001||International Business Machines Corporation||Selective tunneling of streaming data|
|US6195622||Jan 15, 1998||Feb 27, 2001||Microsoft Corporation||Methods and apparatus for building attribute transition probability models for use in pre-fetching resources|
|US6223226||Jul 7, 1998||Apr 24, 2001||Mitsubishi Denki Kabushiki||Data distribution system and method for distributing data to a destination using a distribution device having a lowest distribution cost associated therewith|
|US6233571||May 4, 1998||May 15, 2001||Daniel Egger||Method and apparatus for indexing, searching and displaying data|
|US6243091||Nov 21, 1997||Jun 5, 2001||International Business Machines Corporation||Global history view|
|US6263364||Nov 2, 1999||Jul 17, 2001||Alta Vista Company||Web crawler system using plurality of parallel priority level queues having distinct associated download priority levels for prioritizing document downloading and maintaining document freshness|
|US6278992||Feb 17, 1999||Aug 21, 2001||John Andrew Curtis||Search engine using indexing method for storing and retrieving data|
|US6285999||Jan 9, 1998||Sep 4, 2001||The Board Of Trustees Of The Leland Stanford Junior University||Method for node ranking in a linked database|
|US6292880||Apr 15, 1998||Sep 18, 2001||Inktomi Corporation||Alias-free content-indexed object cache|
|US6336123||Oct 2, 1997||Jan 1, 2002||Matsushita Electric Industrial Co., Ltd.||Hierarchical based hyper-text document preparing and management apparatus|
|US6366947||Jan 20, 1998||Apr 2, 2002||Redmond Venture, Inc.||System and method for accelerating network interaction|
|US6370580||Sep 15, 1999||Apr 9, 2002||Clearway Acquisition, Inc.||Web serving system that coordinates multiple servers to optimize file transfers|
|US6374336||Apr 3, 1998||Apr 16, 2002||Avid Technology, Inc.||Computer system and process for transferring multiple high bandwidth streams of data between multiple storage units and multiple applications in a scalable and reliable manner|
|US6404446||Feb 22, 2000||Jun 11, 2002||International Business Machines Corporation||Multi-node user interface component and method thereof for use in displaying visual indication of search results|
|US6418452||Nov 3, 1999||Jul 9, 2002||International Business Machines Corporation||Network repository service directory for efficient web crawling|
|US6427172||Jul 20, 1999||Jul 30, 2002||Cyberstar, L.P.||Method and apparatus for internet cache content delivery via a data distribution system|
|US6467046||May 6, 1996||Oct 15, 2002||Sun Microsystems, Inc.||System and method for automatically distributing copies of a replicated database in a computer system|
|US6477548||Sep 27, 1999||Nov 5, 2002||Casio Computer Co., Ltd.||Server computer capable of automatically notifying updated database content to client computer, and program storage medium|
|US6477566||Jan 21, 1998||Nov 5, 2002||Nortel Networks Limited||Method and system of providing improved network management data between a plurality of network elements and a management system for increasing a flow and decreasing an amount of data transfer|
|US6484204||May 24, 2001||Nov 19, 2002||At&T Corp.||System and method for allocating requests for objects and managing replicas of objects on a network|
|US6499052||Aug 11, 1999||Dec 24, 2002||Yahoo! Inc.||Electronic commerce system for referencing remote commerce sites at a local commerce site|
|US6547829||Jun 30, 1999||Apr 15, 2003||Microsoft Corporation||Method and system for detecting duplicate documents in web crawls|
|US6606643||Jan 4, 2000||Aug 12, 2003||International Business Machines Corporation||Method of automatically selecting a mirror server for web-based client-host interaction|
|US6618752||Apr 18, 2000||Sep 9, 2003||International Business Machines Corporation||Software and method for multicasting on a network|
|US6631369 *||Jun 30, 1999||Oct 7, 2003||Microsoft Corporation||Method and system for incremental web crawling|
|US6665658||Jan 13, 2000||Dec 16, 2003||International Business Machines Corporation||System and method for automatically gathering dynamic content and resources on the world wide web by stimulating user interaction and managing session information|
|US6671705||Aug 17, 1999||Dec 30, 2003||Emc Corporation||Remote mirroring system, device, and method|
|US6675159||Jul 27, 2000||Jan 6, 2004||Science Applic Int Corp||Concept-based search and retrieval system|
|US6687696||Jul 26, 2001||Feb 3, 2004||Recommind Inc.||System and method for personalized search, information filtering, and for generating recommendations utilizing statistical latent class models|
|US6711568||Nov 8, 2000||Mar 23, 2004||Krishna Asur Bharat||Method for estimating coverage of web search engines|
|US6718361||Apr 7, 2000||Apr 6, 2004||Network Appliance Inc.||Method and apparatus for reliable and scalable distribution of data files in distributed networks|
|US6725269||Dec 2, 1999||Apr 20, 2004||International Business Machines Corporation||System and method for maintaining multiple identities and reputations for internet interactions|
|US6826576 *||Sep 25, 2001||Nov 30, 2004||Microsoft Corporation||Very-large-scale automatic categorizer for web content|
|US6847967||Aug 18, 2000||Jan 25, 2005||Nec Corporation||Information set importance determination system and information set importance determination method|
|US6859839||Aug 7, 2000||Feb 22, 2005||Wisconsin Alumni Research Foundation||Bandwidth reduction of on-demand streaming data using flexible merger hierarchies|
|US6901604||Apr 27, 1999||May 31, 2005||Chaincast, Inc.||Method and system for ensuring continuous data flow between re-transmitters within a chaincast communication system|
|US6947930||Apr 25, 2003||Sep 20, 2005||Overture Services, Inc.||Systems and methods for interactive search query refinement|
|US6970937||Jun 15, 2001||Nov 29, 2005||Abacast, Inc.||User-relayed data broadcasting|
|US6976207||Apr 27, 2000||Dec 13, 2005||Ser Solutions, Inc.||Classification method and apparatus|
|US6978419||Nov 15, 2000||Dec 20, 2005||Justsystem Corporation||Method and apparatus for efficient identification of duplicate and near-duplicate documents and text spans using high-discriminability text fragments|
|US6990628 *||Jun 14, 1999||Jan 24, 2006||Yahoo! Inc.||Method and apparatus for measuring similarity among electronic documents|
|US6990667||Jan 29, 2002||Jan 24, 2006||Adaptec, Inc.||Server-independent object positioning for load balancing drives and servers|
|US6996668||Jun 28, 2002||Feb 7, 2006||Seagate Technology Llc||Synchronized mirrored data in a data storage device|
|US7035933||Sep 13, 2001||Apr 25, 2006||Network Foundation Technologies, Inc.||System of distributing content data over a computer network and method of arranging nodes for distribution of data over a computer network|
|US7039173||Jan 8, 2002||May 2, 2006||Nokia Corporation||Management of performance of intelligent network services|
|US7065618||Jun 30, 2003||Jun 20, 2006||Google Inc.||Leasing scheme for data-modifying operations|
|US7080073||Jun 24, 2002||Jul 18, 2006||Firstrain, Inc.||Method and apparatus for focused crawling|
|US7325073||Jun 4, 2003||Jan 29, 2008||International Business Machines Corporation||Peer to peer video on demand system selectively using client and server to forward the requested video program to another client|
|US20020023158||Apr 27, 2001||Feb 21, 2002||Polizzi Kathleen Riddell||Method and apparatus for implementing search and channel features in an enterprise-wide computer system|
|US20020038350||Apr 30, 2001||Mar 28, 2002||Inceptor, Inc.||Method & system for enhanced web page delivery|
|US20020057678||Aug 16, 2001||May 16, 2002||Jiang Yuen Jun||Method and system for wireless voice channel/data channel integration|
|US20020059416||Jan 8, 2002||May 16, 2002||Heikki Tuunanen||Management of performance of intelligent network services|
|US20020059592||May 17, 2001||May 16, 2002||Chaincast, Inc.||Method and system for implementing an internet radio device for receiving and /or transmitting media information|
|US20020078014||May 30, 2001||Jun 20, 2002||David Pallmann||Network crawling with lateral link handling|
|US20020078035||Feb 22, 2001||Jun 20, 2002||Frank John R.||Spatially coding and displaying information|
|US20020087599 *||Feb 22, 2002||Jul 4, 2002||Grant Lee H.||Method of coding, categorizing, and retrieving network pages and sites|
|US20020103809||Feb 2, 2001||Aug 1, 2002||Searchlogic.Com Corporation||Combinatorial query generating system and method|
|US20020103968||Jan 31, 2001||Aug 1, 2002||Grover Rajiv K.||Mirroring agent accessible to remote host computers, and accessing remote data-storage devices, via a communcations medium|
|US20020120685||Apr 10, 2002||Aug 29, 2002||Alok Srivastava||System for dynamically invoking remote network services using service descriptions stored in a service registry|
|US20020138509||Jan 18, 2001||Sep 26, 2002||Michael Burrows||System and method for storing connectivity information in a web database|
|US20020138559||Jan 29, 2002||Sep 26, 2002||Ulrich Thomas R.||Dynamically distributed file system|
|US20020169818||Apr 24, 2002||Nov 14, 2002||Compaq Computer Corporation||Method and apparatus for efficient storage and retrieval of objects in and from an object storage device|
|US20020194324||Apr 26, 2002||Dec 19, 2002||Aloke Guha||System for global and local data resource management for service guarantees|
|US20030014399||Mar 12, 2002||Jan 16, 2003||Hansen Mark H.||Method for organizing records of database search activity by topical relevance|
|US20030014523||Jul 13, 2001||Jan 16, 2003||John Teloh||Storage network data replicator|
|US20030028726||Jun 28, 2002||Feb 6, 2003||Gaertner Mark A.||Synchronized mirrored data in a data storage device|
|US20030037131||Aug 17, 2001||Feb 20, 2003||International Business Machines Corporation||User information coordination across multiple domains|
|US20030046283||Oct 29, 2002||Mar 6, 2003||Gracenote Inc.||Method and system for finding approximate matches in database|
|US20030101278||Sep 10, 2002||May 29, 2003||J.J. Garcia-Luna-Aceves||System and method for directing clients to optimal servers in computer networks|
|US20030130994||Sep 26, 2002||Jul 10, 2003||Contentscan, Inc.||Method, system, and software for retrieving information based on front and back matter data|
|US20030131193||Feb 24, 2003||Jul 10, 2003||Hitachi, Ltd.||Multiple processor data processing system with mirrored data for distributed access|
|US20030195883||Apr 15, 2002||Oct 16, 2003||International Business Machines Corporation||System and method for measuring image similarity based on semantic meaning|
|US20030200307||Sep 10, 2002||Oct 23, 2003||Jyoti Raju||System and method for information object routing in computer networks|
|US20030204571||Apr 24, 2002||Oct 30, 2003||International Business Machines Corporation||Distributed file system using scatter-gather|
|US20030212869||May 5, 2003||Nov 13, 2003||Burkey Todd R.||Method and apparatus for mirroring data stored in a mass storage system|
|US20030225884||May 31, 2002||Dec 4, 2003||Hayden Mark G.||Distributed network storage system with virtualization|
|US20040019721||Jul 25, 2002||Jan 29, 2004||International Business Machines Corporation||Data path master/slave data processing device apparatus and method|
|US20040049673||Sep 5, 2002||Mar 11, 2004||Docomo Communications Laboratories Usa, Inc.||Apparatus and method for a personal cookie repository service for cookie management among multiple devices|
|US20040093361||Sep 10, 2003||May 13, 2004||Therrien David G.||Method and apparatus for storage system to provide distributed data storage and protection|
|US20040128383||Dec 31, 2002||Jul 1, 2004||International Business Machines Corporation||Method and system for enroll-thru operations and reprioritization operations in a federated environment|
|US20040143672||Jan 7, 2003||Jul 22, 2004||Microsoft Corporation||System and method for distributing streaming content through cooperative networking|
|US20040205149||Sep 10, 2003||Oct 14, 2004||Hughes Electronics||System and method for pre-fetching content in a proxy architecture|
|US20040210575||Apr 18, 2003||Oct 21, 2004||Bean Douglas M.||Systems and methods for eliminating duplicate documents|
|US20040223615||Apr 9, 2004||Nov 11, 2004||Dhawan Atam P.||Methods and apparatus for multi-level dynamic security system|
|US20040260873||Jun 17, 2003||Dec 23, 2004||Hitachi, Ltd.||Method and apparatus for managing replication volumes|
|US20050015436||May 9, 2003||Jan 20, 2005||Singh Ram P.||Architecture for partition computation and propagation of changes in data replication|
|US20050027685||Aug 23, 2004||Feb 3, 2005||Kamvar Sepandar D.||Adaptive computation of ranking|
|US20050154887||Jan 12, 2004||Jul 14, 2005||International Business Machines Corporation||System and method for secure network state management and single sign-on|
|US20050165778||Dec 22, 2004||Jul 28, 2005||Microsoft Corporation||Adaptive Web crawling using a statistical model|
|US20050216845||Oct 29, 2004||Sep 29, 2005||Jason Wiener||Utilizing cookies by a search engine robot for document retrieval|
|US20050273514||May 9, 2001||Dec 8, 2005||Ray Milkey||System and method for automated and optimized file transfers among devices in a network|
|US20060129775||Feb 10, 2006||Jun 15, 2006||Hulsey John A||Method, system and apparatus for releasing storage in a fast replication environment|
|US20060271705||Aug 3, 2006||Nov 30, 2006||Garcia-Luna-Aceves J J||System and method for discovering information objects and information object repositories in computer networks|
|US20070174471||Apr 28, 2004||Jul 26, 2007||Cedric Van Rossum||Secure, continous, proxy-optimized, device-to-device data download reception system and method of use|
|US20080005780||Sep 14, 2007||Jan 3, 2008||Singleton Richard B||Master security policy server|
|US20080016028||Jul 19, 2007||Jan 17, 2008||International Business Machines Corporation||Dynamic Cluster Database Architecture|
|USD295632||Oct 28, 1985||May 10, 1988||Xerox Corporation||Icon for wastebasket or the like|
|USD295764||Dec 9, 1985||May 17, 1988||Xerox Corporation||Icon for telephone or the like|
|USD396455||Jan 21, 1994||Jul 28, 1998||Xerox Corporation||Display object for a display screen|
|USD469444||Jan 4, 2000||Jan 28, 2003||Apple Computer, Inc.||User interface for computer display|
|USD472244||Apr 19, 2002||Mar 25, 2003||Apple Computer, Inc.||User interface for computer display|
|USD473565||Mar 3, 2000||Apr 22, 2003||Cognos Incorporated||User interface for a computer screen|
|USD474201||Apr 5, 2002||May 6, 2003||Apple Computer, Inc.||User interface for computer display|
|1||Adam, N. et al., "Regeneration with Virtual Copies for Distributed Computing Systems," IEEE Transactions on Software Engineering, vol. 19, No. 6, Jun. 1993, 9 pages.|
|2||Alvarez, M. et al., "Client-Side Deep Web Data Extraction," IEEE International Conference on E-Commerce Technology for Dynamic E-Business, Sep. 13-15, 2004, 5 pages.|
|3||Bawa, M.et al., "Transience of Peers and Streaming Media," ACM SIGCOMM Computer Communications Review, vol. 33, No. 1, Jan. 2003, 6 pages.|
|4||Bharat, K. et al., "A Comparison of Techniques to Find Mirrored Hosts on the WWW," Proceedings Workshop on Organizing Web Space at 4th ACM Conference on Digital Libraries, Aug. 25, 1999, 19 pages.|
|5||Bharat, K. et al., "Mirror, Mirror on the Web: A Study of Host Pairs with Replicated Content," In the Proceedings of the Eighth International World Wide Web Conference, 1999, 17 pages.|
|6||Brin, S. et al., "Copy Detection Mechanisms for Digital Documents," Proceedings of the ACM SIGMOD International Conference on Management of Data, Oct. 31, 1994, 21 pages.|
|7||Brin, S. et al., "The Anatomy of a Large-Scale Hypertextual Web Search Engine," Computer Networks and ISDN Systems, 1998, 11 pages.|
|8||Broder, A.Z., "On Resemblance and Containment of Documents," Proceedings of Compression and Complexity of Sequences, IEEE Computer Society, 1997, 9 pages.|
|9||Brusilovsky, P., et al, "Map-Based Horizontal Navigation in Educational Hypertext," HT'02, Jun. 11-15, 2002, 10 pages.|
|10||Cerf, V. et al., "Specification of Internet Transmission Control Program," RFC 675, Specification of Internet Transmission Control Program, Dec. 1974, 141 pages.|
|11||Chawathe, Y. et al., "An Architecture for Internet Content Distribution as an Infrastructure Service," http://www.cs.berkeley.edu/~yatin/papers/scatterca, 2000, 16 pages.|
|12||Chawathe, Y. et al., "An Architecture for Internet Content Distribution as an Infrastructure Service," http://www.cs.berkeley.edu/˜yatin/papers/scatterca, 2000, 16 pages.|
|13||Chawathe, Y., "Scattercast: An Architecture for Internet Broadcast Distribution as an Infrastructure Service," Dissertation of Yatin Dilip Chawathe, University of California at Berkeley, Fall 2000, 161 pages.|
|14||Cho, J., et al., "Finding Replicated Web Collections," Proceedings of the ACM SIGMOD International Conference on Management of Data, 25 pages, 2000.|
|15||Chowdhury, A., et al., "Collection Statistics for Fast Duplicate Document Detection," ACM Transactions on Information Systems, vol. 20, No. 2, Apr. 2002, 21 pages.|
|16||Cohen, B., Incentives Build Robustness in Bit Torrent, firstname.lastname@example.org, May 22, 2003, 5 pages.|
|17||*||David B. Leake, towards context-based search engine selection, 2001, ACM Press, pp. 109-112.|
|18||*||Filippo Menczer, Evaluating topic-driven web crawlers, 2001, ACM Press, pp. 241-249.|
|19||Haveliwala, T., "Topic-Sensitive Page Rank," WWW2002, May 7-11, 2002, 10 pages.|
|20||Henzinger, M. et al., "Challenges in Web Search Engines," Internet Mathematics, vol. 1, No. 1, Oct. 17, 2002, 14 pages.|
|21||Ipeirotis, P. et al., "Extending SDARTS: Extracting Metadata from Web Databases and Interfacing with the Open Archives Initiative," Proceedings of the 2nd ACM/IEEE-CS Joint Conference on Digital Libraries Table of Contents, 2002, 9 pages.|
|22||Jeh, G. et al., "Scaling Personalized Web Search,"WWW2003, May 20-24, 2003, 9 pages.|
|23||Kamvar, S. et al., "Exploiting the Block Structure of the Web for Computing Page Rank," Stanford University Technical Report, Jun. 2003, 14 pages.|
|24||Kelly, T. et al., "Aliasing on the World Wide Web: Prevalence and Performance Implications," Proceedings of the 11th International World Wide Web Conference, May 7-11, 2002, 11 pages.|
|25||Kleinberg, J., "Authoritative Sources in a Hyperlinked Environment," Journal of the ACM, vol. 46, No. 5, Sep. 1999, 29 pages.|
|26||Koman, R., "How the Wayback Machine Works," XML.com, Jan. 21, 2001, 6 pages.|
|27||Kristol, D. et al., "HTTP State Management Mechanism," RFC 2965, HttP Saae Manlagement Mechanism, Oct. 2000, 25 pages.|
|28||Leuski, A., "Evaluating Document Clustering for Interactive Information Retrieval," Proceedings of the 10th Int'l Conference on Information and Knowledge, 2001, 13 pages.|
|29||Miller, R., "WebSPHINX: A Personal, Customizable Web Crawler," http://www.cs.cmu.edu/~rcm/websphinx, Jul. 8, 2002, 10 pages.|
|30||Miller, R., "WebSPHINX: A Personal, Customizable Web Crawler," http://www.cs.cmu.edu/˜rcm/websphinx, Jul. 8, 2002, 10 pages.|
|31||Nguyen, T. et al., "Distributed Video Streaming Over Internet," Proc. SPIE, vol. 4673, 2002, 10 pages.|
|32||Notess, G., "The Wayback Machine: The Web's Archive,"Online, vol. 26, No. 2, Mar./Apr. 2002, 4 pages.|
|33||Osokine, S., "Search Optimization in the Distributed Networks," http://www.grouter.net/gnutella/search.lhtm, Oct. 15, 2002, 65 pages.|
|34||Padmanabhan, V. et al., "Distributing Streaming Media Content Using Cooperative Networking," NOSSDAV '02, May 12-14, 2002, 10 pages.|
|35||Postel, J. et al., "File Transfer Protocol (FTP)," Obsoletes RFC 765 (IEN 149) Oct. 1985, 69 pages.|
|36||Raghavan, S. et al., "Crawling the Hidden Web," Technical Report, Stanford, Dec. 7, 2000, 26 pages.|
|37||Rowstron, A. et al., "Storage Management and Caching in PAST, a Large-Scale, Persistent Peer-to-Peer Storage Utility," SOSP-18, Nov. 2001, 13 pages.|
|38||Sarawagi, S., et al., "Interactive Deduplication Using Active Learning," Proceeding of the 8th ACM SIGKDD Int'l Conf. on Knowledge, Discovery and Data Mining, Jul. 2002, 10 pages.|
|39||Shivakumar, N., et al., "Finding Near-Replicas of Documents on the Web," in World Wide Web and Databases, International Workshop WebDB'98, Mar. 1998, 4 pages.|
|40||Smith, B. et al., "Exploiting Result Equivalence in Caching Dynamic Web Content," USENIX Symposium on Internet Technology and Systems, Oct. 1999. 12 pages.|
|41||Stoica, I. et al., "Chord: A Scalable Peer-to-Peer Lookup Service for Internet Applications," MIT Laboratory for Computer Science, Jan. 10, 2002, 45 pages.|
|42||Tsoi, "Structure of the Internet?" Faculty of Informatics-Papers, Univ. of Wollongong, 2001, 5 pages.|
|43||Wang, J. et al., "Web Search Services," Univ. of Science & Technology, Hong Kong, Computer Science Technical Report, HKUST-CS02-26, Jan. 1999, 35 pages.|
|44||Wolfson O. et al., "An Adaptive Data Replication Algorithm," ACM Transactions on Database Systems, vol. 22, No. 2, Jun. 1997, 60 pages.|
|45||Wolfson O. et al., "Distributed Algorithms for Dynamic Replication of Data," 11th Principles of Database Systems, Jun. 1992, 15 pages.|
|46||Zhao, B. et al., Tapestry: An Infrastructure for Fault-Tolerant Wide-Area Location and Routing, Computer Science Division (EECS), University of California, Berkeley, Apr. 2001, 28 pages.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8307076 *||Nov 3, 2010||Nov 6, 2012||Google Inc.||Content retrieval from sites that use session identifiers|
|US8375066 *||Apr 26, 2010||Feb 12, 2013||International Business Machines Corporation||Generating unique identifiers|
|US8407186 *||Mar 31, 2009||Mar 26, 2013||Symantec Corporation||Systems and methods for data-selection-specific data deduplication|
|US8560698||Jun 27, 2010||Oct 15, 2013||International Business Machines Corporation||Allocating unique identifiers using metadata|
|US8713009 *||Sep 25, 2008||Apr 29, 2014||Yahoo! Inc.||Associating objects in databases by rate-based tagging|
|US8762338 *||Feb 12, 2010||Jun 24, 2014||Symantec Corporation||Analyzing backup objects maintained by a de-duplication storage system|
|US8918392||Mar 29, 2012||Dec 23, 2014||Amazon Technologies, Inc.||Data storage mapping and management|
|US8930364 *||Mar 29, 2012||Jan 6, 2015||Amazon Technologies, Inc.||Intelligent data integration|
|US8935203||Mar 29, 2012||Jan 13, 2015||Amazon Technologies, Inc.||Environment-sensitive distributed data management|
|US9044543||Jul 17, 2012||Jun 2, 2015||Elwha Llc||Unmanned device utilization methods and systems|
|US9061102||Aug 31, 2012||Jun 23, 2015||Elwha Llc||Unmanned device interaction methods and systems|
|US9118571||Jul 8, 2013||Aug 25, 2015||Telefonaktiebolaget L M Ericsson (Publ)||Methods of operating load balancing switches and controllers using matching patterns with unrestricted characters|
|US9137165||Jun 17, 2013||Sep 15, 2015||Telefonaktiebolaget L M Ericsson (Publ)||Methods of load balancing using primary and stand-by addresses and related load balancers and servers|
|US9254363||Aug 31, 2012||Feb 9, 2016||Elwha Llc||Unmanned device interaction methods and systems|
|US9456030||Sep 15, 2014||Sep 27, 2016||Telefonaktiebolaget Lm Ericsson (Publ)||Methods of operating load balancing switches and controllers using modified flow entries|
|US9485183||Apr 25, 2014||Nov 1, 2016||Telefonaktiebolaget Lm Ericsson (Publ)||System and method for efectuating packet distribution among servers in a network|
|US9531809||Aug 11, 2014||Dec 27, 2016||Amazon Technologies, Inc.||Distributed data storage controller|
|US9575680||Aug 22, 2014||Feb 21, 2017||Veritas Technologies Llc||Deduplication rehydration|
|US9621642||Jun 17, 2013||Apr 11, 2017||Telefonaktiebolaget Lm Ericsson (Publ)||Methods of forwarding data packets using transient tables and related load balancers|
|US9713675||Aug 31, 2012||Jul 25, 2017||Elwha Llc||Unmanned device interaction methods and systems|
|US9733644||Aug 31, 2012||Aug 15, 2017||Elwha Llc||Unmanned device interaction methods and systems|
|US9798325||Aug 31, 2012||Oct 24, 2017||Elwha Llc||Unmanned device interaction methods and systems|
|US20090216723 *||Jan 9, 2009||Aug 27, 2009||Computer Associates Think, Inc.||Directory Partitioned System and Method|
|US20100082576 *||Sep 25, 2008||Apr 1, 2010||Walker Hubert M||Associating objects in databases by rate-based tagging|
|US20110082841 *||Feb 12, 2010||Apr 7, 2011||Mark Christiaens||Analyzing Backup Objects Maintained by a De-Duplication Storage System|
|US20110264706 *||Apr 26, 2010||Oct 27, 2011||International Business Machines Corporation||Generating unique identifiers|
|US20140082183 *||Sep 9, 2013||Mar 20, 2014||Salesforce.Com, Inc.||Detection and handling of aggregated online content using characterizing signatures of content items|
|US20140372616 *||Jun 17, 2013||Dec 18, 2014||Telefonaktiebolaget L M Ericsson (Publ)||Methods of forwarding/receiving data packets using unicast and/or multicast communications and related load balancers and servers|
|US20170103219 *||Dec 20, 2016||Apr 13, 2017||Imdb.Com, Inc.||Restricting network spidering|
|WO2013173802A1 *||May 17, 2013||Nov 21, 2013||Google Inc.||Systems and methods for crawling and indexing content|
|WO2013173805A1 *||May 17, 2013||Nov 21, 2013||Google Inc.||Systems and methods re-ranking ranked search results|
|U.S. Classification||715/200, 707/634, 707/637, 707/740|
|Cooperative Classification||G06F17/30112, G06F17/30864, H04L63/08, H04L41/22, G06F17/3053, H04L29/06, H04L67/2804, H04L63/102, G06F17/30867, H04L29/0809, H04L67/14, G06F17/303, H04L41/12, G06Q30/0246, H04L29/08072, G06Q30/02|
|Sep 8, 2004||AS||Assignment|
Owner name: GOOGLE INC., CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:ZHU, HUICAN;ACHARYA, ANURAG;SIGNING DATES FROM 20040324 TO 20040330;REEL/FRAME:015094/0581
|Sep 14, 2015||FPAY||Fee payment|
Year of fee payment: 4